Fig. 4: Drought-induced wheat yield change between irrigated and non-irrigated areas.

According to the location of irrigated areas shown in Fig. 1, the wheat yield change caused by droughts in irrigated areas was calculated. The equal non-irrigated areas around the irrigated areas were then selected to calculate the wheat yield change. The effect of irrigation on reducing agricultural loss due to droughts in different intensities and under different scenarios was assessed and compared in a and b, respectively. The mean value of wheat yield change calculated by Multiple Linear Regression method (MLR), Deep Learning algorithm (DL), and Erosion-Productivity Impact Calculator model (EPIC) was used for the boxplots. The upper, center, and lower lines of each box show the 75^th, 50^th, and 25^th percentiles of the data, respectively. The upper and lower limits mean the third (i.e., 75^th) quartile plus 1.5*Interquartile range (IQR), and the first quartile (i.2., 25^th) minus 1.5*IQR, respectively. IQR is the range from the first quartile to the third quartile. If the upper or lower limit exceeds the maximum or minimum value of the data, then the maximum or minimum value is used as the upper or lower boundary of the error bar.